Conference Agenda

Digital technologies and artifacts are an integral part of primary school children’s everyday lives. This has led to increasing calls to foster competencies for a digitally networked world already at the primary school level. Such competencies are considered central to comprehensive participation in political, cultural, and economic processes within society (Wissenschaftsrat, 2020) and may enable children to actively shape the world in democratic and sustainable ways (GDSU, 2021). Accordingly, competencies for a digitally networked world aim at a self-determined, responsible, and critical use of digital media, as well as at the development of skills for working with, analyzing, reflecting on, and designing digital artifacts (Brinda et al., 2020).

Which specific competencies primary school children should acquire in this context is answered differently depending on the discourse and the underlying reference discipline. Normative and theoretical models as well as empirical studies variously address media, digital, or information literacy. These constructs differ substantially in how they are conceptualized and operationalized (Mensonides et al., 2024). Especially for the primary school level, existing approaches are predominantly normative and have only been insufficiently validated empirically. In recent years, computer science–related competencies have increasingly been addressed within discussions of media, digital, and information literacy, for example in large-scale assessments such as the most recent ICILS study (Eickelmann et al., 2024).

A key aim in this context is to convey foundational computer science concepts underlying the applications that permeate children’s everyday lives, such as intelligent toys (Schmid et al., 2021). This raises the question of which specific computer science competencies should already be initiated in primary education. Given that computer science as a reference discipline encompasses a broad range of subfields, a wide variety of domains and content areas may be considered, including theoretical computer science (e.g., cryptography or machine models), practical computer science (e.g., data structures and algorithms), applied computer science (e.g., databases or operating systems), and computer engineering (e.g., robotics or networks).

Various computer science associations have derived normative competence descriptions for primary education that include multiple subdimensions. In Germany, the Informatics Society (2019) specifies five contet areas for the end of grades two and four: Information and Data, Algorithms, Languages and Automata, Computing Systems, and Computing, Humans, and Society. Similar domains are defined in the K–12 Computer Science Standards (Computer Science Teachers Association, 2017) for children aged 5–7 and 8–11, complemented by the domain Networks and the Internet. Within each domain, further subconcepts are distinguished, such as devices, hardware and software, or troubleshooting. In England, the computing curriculum likewise formulates learning objectives across all school levels. In addition to domains such as Computing Systems and Networks, it includes areas such as Creating Media, which are often also associated with media- or digital-related competencies (e.g., DigComp 2.2, Vuorikari et al., 2022).

This exemplary overview of curricula and frameworks highlights that terminology and conceptualizations vary considerably and reflect different disciplinary perspectives and content emphases. At the same time, different aspects of literacy are foregrounded, such as knowledge, skills, or attitudes. Against this background, the overarching question arises of how existing empirical findings and constructs can be systematically synthesized to provide a comprehensive overview of computer science competencies for primary education. The present study addresses the following research questions:

a) How is the construct of computer science competencies conceptualized at the primary school level, and how is it distinguished from related constructs?

b) Which subdimensions of computer science competencies can be identified, and how are they specified for the primary school age group?

Numerous studies conducted over the past decades have shown that girls tend to be less interested in computers, use them less often in their spare time, and have a more negative attitude toward computers (Brosnan, 1998; Cooper, 2006; Tondeur et al., 2016). Consequently, it has long been assumed that they are often less computer literate than boys. However, recent studies suggest that gender differences in information- and communication technology (ICT)-related skills need to be viewed in a more nuanced way.

The results of performance-based ICT tests must be distinguished from measurements based on self-assessment. Performance-based tests in particular often show no clear advantage for boys. In various recent studies that assess ICT literacy using behavior-based methods or knowledge-based tests, girls outperform boys or negligible gender differences were found (Gebhardt et al., 2019; Gnambs, 2021; Qazi et al., 2022). These findings are corroborated by a meta-analysis by Siddiq and Scherer (2019), who found a significant positive difference in favor of girls in performance-based tests. In contrast, test procedures based on self-assessment are also mixed but tend to favor boys (Christensen, 2023). A reason for this might be that these procedures do not measure actual behavioral skills or knowledge, but rather skill-related beliefs (Scherer & Siddiq, 2019). This assumption is also supported by studies that measure self-reported beliefs such as ICT-related self-efficacy or self-concept. For example, Gnambs (2021) uses longitudinal data from an education panel survey to show that boys report higher ICT-related self-efficacy or self-perception than girls over time – despite not having any corresponding advantages in terms of actually measured ICT literacy.

The differences reported are particularly relevant in a school context, given that educational institutions are mandated to promote ICT literacy among all genders. This includes imparting declarative and conditional knowledge as well as providing learning opportunities for the practical application and testing of these skills. In this way, the corresponding self-efficacy in dealing with digital technologies can and should be strengthened at the same time.

Until now, gender differences in performance-based ICT tests, ICT-related competence beliefs, and ICT use in schools have rarely been considered in conjunction with one another. Gerick et al. (2019) report that the performance gap between boys and girls in a performance-based ICT literacy test can be explained in part by their ICT-related self-efficacy beliefs, while keeping the use of digital media in schools constant. Different effects were observed between beliefs regarding basic and advanced skills. In a reanalysis of ICILS data from 2013 and 2018, Campos and Scherer (2024) confirmed that girls with equal ICT access to boys have better digital skills, partly due to differences in ICT self-efficacy in basic computer skills. However, they did not find the same association in items assessing students' ICT self-efficacy in advanced skills. In a study of elementary school students, Aesart et al. (2017) found that girls assessed their ICT literacy more accurately than boys, but the gender difference in the accuracy of skill assessments disappeared when actual performance in a skill-based test was included in the model.

The the present study aims to analyze the relationship between computer-related self-efficacy beliefs and ICT literacy more closely, taking into account gender and school computer use. In addition to the global view of ICT literacy, individual skill facets (informing, producing, communicating) are also considered.

In e-learning, the development of AI-based technologies is creating new expectations regarding the potential for adapting training depending on the learner's needs. Studies show that the use of artificial intelligence in online courses can reduce dropout rates (Wan & Yu, 2023), whether through predicting potential dropouts (Khan, 2021; Zerkouk et al. 2023), through the use of adaptive learning (Papamitsiou et al. 2020), or through providing closer guidance via intelligent tutors (Minne, 2025). Since 2015, international publications on IA-supported education have become increasingly important. Jing (2023) points out that a large proportion of these papers come from the field of computer science and focus more on tools and adaptive learning. However, one of the challenges is to analyse the use of AI in training and its potential effects on dropout rates by drawing on theoretical frameworks from the fields of education and training. Several studies highlight a dropout rate that can be high in e-learning environments, particularly among adults, although this depends on the type of training program (Breslow et al., 2013).

In this context, the purpose of this paper is to analyse the extent to which AI could improve learner completion in adult e-learning programs by identifying the different types of factors relating to the learner, his environment and the training program. This question leads us, first of all, to consider the use of other types of AI, such as generative AI; then, to pay particular attention to instructional design (Paquelin, 2020), based on the theoretical fundamentals of adult learning. The literature emphasises that one of the important issues for adult learners is the motivational aspect, which must be taken into consideration. Numerous theories have been developed in the literature on motivation (Fenouillet, 2012), which is a complex and multidimensional concept that requires contextual factors to be taken into account (Aubret, 2008). Adult learners have financial, family and professional responsibilities that can affect their available time (Njingang et al., 2021; Papi et al., 2022), the nature of their motivation (Deci & Ryan, 1985; Kember, 1989), their goals (Kizilcec et al., 2017, Carré, 2022; 2024) and their self-efficacy (Bandura; Vayre & Vonthron, 2018).

Based on this literature, a set of factors related to individual and contextual characteristics of the learners as well as the training program itself have been identified as potential predictors of adult learners' dropout. Through an empirical study of an e-learning program for adults, we first seek to evaluate and determine the predictive factors of dropout. Secondly, the model will be enhanced by the integration of generative AI and, more specifically, motivational messages generated and customized to the learner's profile (Mélot et al., 2017).

Schools are increasingly integrating data-driven digital technologies such as learning analytics into everyday teaching. These processes are contributing to the broader evolution called the “datafication” of education, which is reshaping educational practices, processes, and even values sometimes in disputed ways (Eynon, 2022). While data platforms promise speed, efficiency, and objectivity in education by generating indicators of student performance for instructional decision-making , they also introduce new forms of “knowing” through dashboards and metrics, thus also altering relational dynamics between teachers and students (see e.g., Jarke & Macgilchrist, 2021; Selwyn et al., 2023). Following Mertala (2024), we conceptualise this teacher-student-data relationship sociomaterially as a triadic relational field, where data-driven representations act as a “third participant” that shapes the conditions of and possibilities for dialogic presence between human actors. Although data are often framed as neutral aids to inform decisions, they carry implicit and powerful assumptions about what counts as valuable learning and influence how teachers and students act on such information. This raises tensions between platform-driven data logics and the aspiration to sustain well-rounded, dialogic educatio n, which warrant deeper scholarly attention.

This work-in-progress paper aims to acquire in-depth knowledge of the empirical conditions, potentials , and pitfalls of data-driven educational relations at the grassroots-level of schools. It examines how the digital platform “ViLLE” (see e.g., Kurvinen et al., 2020) mediates teacher-student interactions in the classroom and what this means for educational purposes and ethics. ViLLE is a compelling example of the potential power of datafication. It collects and visualises data in real-time drill-and-practice activities for mathematics learning and provides teachers and students with automatically generated metrics through visually rich dashboards. These visualisations automatically highlight (and even interpret) students’ performance, progress, and areas of difficulty . The preliminary research questions are:

(1) How do teacher-student interactions mediated by ViLLE’s data representations reflect what is considered valuable in education?

(2) How do these interactions unfold in terms of dialogic presence and mutuality?

The analysis for this two-pronged RQ is underpinned by a dual theoretical heuristic: one that combines Gert Biesta’s (2009; 2022) three domains of educational purpose (qualification, socialisation, and subjectification), which account for well-rounded education, with Martin Buber’s (1958) classic and widely employed relational concepts of I-It and I-Thou (for dual use of these theories, see also Mertala, 2024). The first lens addresses what the educational encounter seeks to achieve: whether it emphasises skill acquisition, the transmission of social norms, or the empowerment of students as autonomous, responsible subjects. It focuses on how such educational purposes are in different ways communicated as being pursued during educational encounters. The second lens, in turn, examines how these encounters unfold relationally: whether the interaction adopts an instrumental stance (I-It) or a dialogic, reciprocal stance (I-Thou), regardless of its pursuit. This novel conceptual approach thus illuminates how data-driven metrics shape both the teleological and the ethical dimensions of educational interactions mediated by data.

The broader aim of the analysis is to identify relational conditions that enable the enactment of dialogic and ethically grounded pedagogy within the context of platform-driven metrics: conditions where teacher, student, and data triad engage without reducing education to, for instance, instrumental logics or undermining dialogic presence (as is commonly criticised in datafication research). By situating these micro-level classroom encounters within ongoing debates on digitalisation and educational accountability, the study intends to contribute to understanding the broader implications of datafication in education.